Security audit
Agent Mistake Patterns 智能体犯错模式库
Security checks for vulnerabilities and agentic risk
Overview
This skill is a disclosed workflow guide for helping agents avoid repeated mistakes, with no executable code or hidden high-impact behavior.
This skill appears safe to install as a process guide. Be aware that it may encourage an agent to consult existing personal notes and write mistake archives, so use it in workspaces where that memory behavior is intended and appropriately scoped.
Vulnerability Patterns
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Static analysis
No suspicious patterns detected.